Every morning at five our delivery performance report is built from the previous day's proof of delivery records, and by eight it is in front of every depot manager. It counts a delivery as on time if the handset recorded a signature or a photograph inside the customer's window.
For most of a year, two of our depots in the hills sat at the bottom of that report, six or seven points behind the rest. One of their managers was put on an improvement plan. He kept saying his drivers were delivering on time, and we kept showing him the report.
He was right. Our handsets store a delivery when there is no signal and send it when there is. On parts of those two depots' routes a driver can go four or five hours without coverage, and some vans come back late with a day's records still waiting on the handset, which only finishes sending once it is docked. Most records arrived before five in the morning. A meaningful share arrived after, and by then the report had already counted them as deliveries that never happened. Customer services, reading the same data at half past eight, were raising a few dozen redelivery requests a week for parcels that were already sitting on doormats.
Nothing was wrong with any individual record. Each one held the correct time of the delivery. What we had never stored was the time it reached us, so nobody could see how much of yesterday was still in transit when the report was frozen.
Every record now carries both times. The report shows, for each depot, what share of the day's expected records had arrived when it was built, and its figures are marked provisional for forty eight hours and final after that. Customer services do not raise a redelivery for a stop while that van's handset is still holding unsent records. And a handset that has sent nothing for six hours raises a ticket for its depot, because by then it is usually a fault rather than a valley.
Once the late records were counted, both depots finished the year above the company average. The improvement plan was withdrawn, and I apologised to the manager in person.
A report with a fixed cut off had been measuring how quickly our data travels as well as how well our drivers worked, and it presented the two as a single number.
– Serguey Shinder
Top comments (0)